The purpose of the study was to analyze NABTEB 2013, 2014 and 2015 Chemistry multiple choice test items for item parameter invariance using item response theory. An evaluation research design was adopted for the study. Unidimensionality test was conducted to see to the IRT statistical assumption being satisfied before further analysis was carried out. The IRT statistical software -eirt – Item Response Theory Assistant for excel was used to determine the item difficulty parameter, item discrimination parameter and guessing parameter estimates. The findings from the study revealed that item difficulty and discrimination parameters were only invariant by sex in 2013 and 2015 but guessing parameters were invariant by sex in 2013, 2014 and 2015. Item difficulty parameters were invariant by school location in 2013, 2014 and 2015 while item discrimination and guessing parameters were invariant by school location in 2014 and 2015 only. Item difficulty and guessing parameters were invariant by school ownership in 2013, 2014 and 2015 but item discrimination parameter was invariant by school ownership in 2013 and 2015. Item difficulty and guessing parameters were invariant by school type in 2013 but item discrimination parameter was invariant by school type in 2013 and 2015. It was recommended that examination bodies such as NABTEB should use any of the subgroups within the population for item analysis provided such subgroups include respondents of the different levels of ability.
Keywords: IRT, Investigation, Item parameter invariance, NABTEB chemistry multiple choice test items, Unidimentionality
